Madagascar’s eight species of endemic carnivores are threatened by deforestation and forest fragmentation, yet relatively little is known about their use of and movement across these degraded landscapes.  The use of local ecological knowledge (LEK) is increasingly recognized as a valuable method for supplementing and guiding scientific studies, while increasing the involvement of local stakeholders in conservation efforts. In October – December 2009, we conducted semi-structured interviews addressing villager experience with endemic and introduced carnivore species with male heads-of-household in communities located 0-20 km from the western edge of the continuous rainforest corridor in southeastern Madagascar. 

Results/Conclusions

Responses from 182 interviews in 17 different communities indicated distinct distributional patterns for two endemic and two introduced species of carnivores, which suggest a range of tolerances to the anthropogenic landscape.  Similar to findings of camera trap surveys conducted in the same region, the largest endemic carnivore, fossa (Cryptoprocta ferox) appeared least tolerant to the human-dominated landscape, and had only been observed in communities ≤ 2.5 km from the continuous forest within the last five years, while the ring-tailed mongoose (Galidia elegans) was found in and around more human-dominated habitats.  Both the introduced Indian civet (Viverricula indica) and wild cats (Felis silvestris) were ubiquitous throughout the study region and displayed an affinity to anthropogenic habitats.  We modeled the probability of observing carnivores within communities using logistic regression and used Akaike’s information criterion adjusted for small sample size (AICc) to determine the best-fitting models for each carnivore species.  Distance to continuous forest was the most important variable for three carnivore species; the probabilities of observing endemic species C. ferox and G. elegans were negatively related to increasing distance to continuous forest, while the opposite trend was observed for the introduced species F. silvestris.  Variables such as the number of households, the proportion of households with dogs, and an index of poultry ownership had lower relative importance in models for these species.  The killing of carnivores, although reported by only a minority of interviewees, was motivated primarily by prevention of or retaliation for poultry depredation, but also for bushmeat, and may be an important mortality risk for species living or moving outside of the continuous forest.  This local ecological knowledge study provided useful information on carnivore distribution and human-carnivore conflict across a large area of human land-use, such as is needed for multi-taxa conservation planning in increasingly fragmented landscapes.